Jun 18, 2026 · 1 hr 6 min · 11 segments
Dan and Brad explore AI as a creative partner for drafting and ideation while requiring strong human direction, and compare today’s shift to the desktop-to-web and mobile transitions, including…
Dan SafferGuestBrad FrostHost
What? But we are at the end of the semester, so I am herding my students along in their final projects, which all happen next week, and they're all using a lot of the latest tools and really wrestling with the capabilities of a lot of new AI things, and they're They are struggling and I completely get it and I understand why.
And yet it's so amazing because there are some amazing things that are happening.
And Claude today updated the model to, it went from, this is where I always blank on the name.
It went from, yeah, it went from Sonnet 4.6 to Opus 4.7, something like that, or was it? Sure, yeah.
So my beef is that And I 100%, I get why they did this, but it is a, I don't know, it's a middle finger to me personally because, so they have this button called adaptive thinking.
And the idea is you switch this adaptive thinking on and it only uses as much processing power as it thinks that you need.
It really is like going back in time to like models, like from like a year, year and a half ago, it is just not that competent.
Rather than if you turn off adaptive thinking, it being very user-friendly and then adapting to just, okay, well, I'll give you the full powered model.
So I'm in the middle of this long writing project, and I've been using 4.6, and it's been going great.
I've just had an amazing time with it, working on this really long writing project that I've been doing.
And now they switched it, and now the model is real janky, it's very user-friendly, and they've clearly done it So that rather than tilt towards knowledge workers, they're narrowly focusing on coders.
Because in... cloud code you can do this thing slash effort where you can you could automatically up the level of thinking for a very long time and and just have it be that permanently you can't do that in the regular cloud chat or cloud projects any of those things so uh so yeah so it's interesting that all of a sudden As of today, or maybe it was yesterday that I launched, whatever, it came out.
What? But we are at the end of the semester, so I am herding my students along in their final projects, which all happen next week, and they're all using a lot of the latest tools and really wrestling with the capabilities of a lot of new AI things, and they're They are struggling and I completely get it and I understand why.
And yet it's so amazing because there are some amazing things that are happening.
And Claude today updated the model to, it went from, this is where I always blank on the name.
It went from, yeah, it went from Sonnet 4.6 to Opus 4.7, something like that, or was it? Sure, yeah.
So my beef is that And I 100%, I get why they did this, but it is a, I don't know, it's a middle finger to me personally because, so they have this button called adaptive thinking.
And the idea is you switch this adaptive thinking on and it only uses as much processing power as it thinks that you need.
It really is like going back in time to like models, like from like a year, year and a half ago, it is just not that competent.
Rather than if you turn off adaptive thinking, it being very user-friendly and then adapting to just, okay, well, I'll give you the full powered model.
So I'm in the middle of this long writing project, and I've been using 4.6, and it's been going great.
I've just had an amazing time with it, working on this really long writing project that I've been doing.
And now they switched it, and now the model is real janky, it's very user-friendly, and they've clearly done it So that rather than tilt towards knowledge workers, they're narrowly focusing on coders.
Because in... cloud code you can do this thing slash effort where you can you could automatically up the level of thinking for a very long time and and just have it be that permanently you can't do that in the regular cloud chat or cloud projects any of those things so uh so yeah so it's interesting that all of a sudden As of today, or maybe it was yesterday that I launched, whatever, it came out.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.